SUMMARY Positron emission tomography (PET) is widely used as a diagnostic molecular imaging technique and clearly has had a significant impact on human healthcare. There have been significant improvements in PET technology by employing time-of-flight (TOF) capability that measures the difference in arrival time of the two 511 keV gamma annihilation rays. This TOF information spatially constrains the origin of the event, leading to improved image quality due to the increased signal-to-noise ratio (SNR) and effective sensitivity in the PET scan. The scintillator bismuth germanate (BGO), however, has several better properties for PET applications than L(Y)SO, such as higher stopping power and higher photo-electric fraction. BGO also has no background activity and can be produced at lower cost than L(Y)SO. However, a major drawback of BGO is the poor coincidence timing resolution originating from the moderate light yield and slow decay time of scintillation photon production. Therefore, BGO has not been considered for use in current generation TOF PET scanners. Theoretically, when a 511keV gamma ray interacts in a scintillator, a small number of optical Cerenkov photons are produced promptly by energetic electrons released by photoelectric or Compton interactions. The Cerenkov photons are produced within an extremely short time frame and earlier than scintillation photons, Thus the Cerenkov photons should provide better timing information for PET. Interestingly, BGO has properties that lead to it producing more Cerenkov photons than other scintillators. Recently, we showed, for the first time, the influence of prompt Cerenkov photons on the timing properties of BGO using latest generation silicon photomultipliers (SiPMs) developed by our partner FBK. Thus, our proposed concept for BGO is that Cerenkov photons are used to obtain prompt timing information while scintillation photons provide essential energy information for PET. The major goals of this proposal are i) to improve coincidence timing resolution of BGO by detecting more Cerenkov photons for TOF PET applications, ii) to develop signal reducing methods and read-out electronics for signals generated by both Cerenkov and scintillation photons, and iii) to finally develop and evaluate the first practical BGO-based TOF PET detector modules, which have 5 5 cm2 cross-sectional area and a coincidence timing resolution of ?300 ps FWHM. If successful, this proposal forms the basis for developing novel BGO-based TOF PET scanners with higher sensitivity and lower cost than current PET scanners.